Update on the Western Weather
Work Group
Carla Thomas
Western IPM Center
Western Plant Diagnostic Network
Western Weather Work Group
•
•
•
•
•
•
•
David Gent, Walt Mahafee, Bill Pfender OSU/ARS
Chris Daly, OSU/Prism
Paul Jepson, Len Coop OSU/IPPC
Gary Grove, Dennis Johnson, Gerrit Hoogenboom, WSU
Carla Thomas, Doug Gubler, Joyce Strand, Neil McRoberts, UCD
Alan Fox, Fox Weather
Emeritus-Fran Pierce WSU, George Taylor OSU
Western Weather Work Group –WIPMC Funding 2005-2010
Mission: To develop a science-based system that
provides principles and procedures to access, synthesize,
distribute, and use weather and climate data products to
improve crop management decision-making abilities
through the delivery of weather based information.
WWWG History
Our vision is to develop access to a backbone network
of physical stations while creating "virtual stations" that
are based on advanced, validated interpolation of
measured variables and model outputs.
• RIPM Proof of Concept funding
• NRI-errors/uncertainties of inputs and outputs of
models
• AFRI-improvement of interpolation, forecasts,
assessments
• PIPE-infrastructure support for existing and emerging
systems
• NPDN-Biosecurity applications/distributed systems
WWWG Research Objectives
• Interpolation of specific variables at necessary time and
space resolutions.
• Development and use of appropriate forecast methods.
• Techniques to estimate difficult-to-measure variables from
other measured variables need to be developed or refined,
and validated.
• Development of standardized modeling structures for
specific types of pathogens to improve availability of
disease models.
• Quantification of uncertainties associated with the various
data and computations so that a level of confidence could
be placed on output and communicated to users.
WWWG Operational Objectives
• Development of networks of weather stations.
• Data acquisition, quality control, storage,
archival, and delivery.
• Focus on needs in accounting for and dealing
with missing data.
• Delivery of pest management applications.
• Training.
• Outreach.
• Evaluation of overall effectiveness.
WWW.USPEST.ORG/WEA
+ 16,000 w
Daily Temperature Regime
Average Temperature
Close alignment
between V2 and
Std.
Over/ Under
estimation of V2
2010 and 2011 at
THILL
Max Temperature
V2 follows
closely with Std
60 in most data
sets.
Over estimation of V2 Max
Temp consistent across
season
Min Temperature
Under prediction
of Min
Temperature is
common in most
Datasets.
Over estimation
of Min Temp is
common at THill.
Temperature means over the data
sets.
Daily Temp
variable
Maximum
Minimum
15 min vs. 60 min interval at
an in-canopy placement
Mean
Mean
Error
absolute
o
o
r
( C)
error ( C)
0.999
-0.21
0.21
0.999
-0.25
0.25
0.998
-0.18
0.27
In-canopy placement vs.
standard placement at 60
min intervals
Mean
Mean
Error
absolute
o
o
r
( C)
error ( C)
0.998
0.42
0.49
0.992
-0.98
0.98
0.993
-0.46
0.82
Grape (WA)
Cherry (WA)
0.999
0.993
Hop (OR)
Hop (WA)
Grape (OR)
0.999
0.998
0.13
0.24
0.13
0.24
0.993
0.987
0.998
0.998
0.12
0.26
0.12
0.26
0.995
0.998
1.00
1.00
1.00
1.00
0.00
0.01
-0.00
0.00
0.05
0.06
0.06
0.06
0.999
0.990
0.998
0.998
Crop (State)
Hop (OR)
Hop (WA)
Grape (OR)
Grape (WA)
Cherry (WA)
Average
Hop (OR)
Hop (WA)
Grape (OR)
Grape (WA)
Cherry (WA)
-0.25
0.25
In-canopy placement at 15
min intervals vs. standard
placement at 60 min
intervals
Mean
Mean
Error
absolute
o
o
r
( C)
error ( C)
0.998
0.21
0.41
0.991 -1.22
1.22
0.992 -0.73
1.02
1.43
0.993 -1.67
Temp
shows
a
good correlation
-0.46
0.50
0.993 -0.33
between
v2
and
-0.60
0.61
0.985 -0.36
Std 60.0.993 0.31
0.19
0.34
-1.42
1.67
Standard placement at 60 min
intervals vs. V2
Mean
Mean
Error
absolute
o
o
r
( C)
error ( C)
0.993
0.59
0.84
0.931
1.22
2.54
0.929
0.17
1.54
Lower Rvalues with
Min2.80Temp
0.965
0.935
1.86
1.61
0.41
0.51
0.958
0.812
0.92
1.48
1.11
2.85
2.20
0.34
0.41
0.996
0.60
0.41
0.63
0.863
0.889
0.779
0.09
1.72
1.00
1.56
2.41
2.49
-0.04
-0.80
-0.17
-0.63
0.18
0.80
0.32
0.63
0.999
0.990
0.998
0.998
-0.04
-0.79
-0.17
-0.63
0.18
0.79
0.32
0.63
0.995
0.941
0.955
0.968
0.943
0.64
0.78
0.15
0.88
0.52
0.68
1.68
1.09
1.33
1.18
Monthly Mean Temp Data
Daily
average
temperature
Silv
March
2010
Silv
April
2010
Silv
May
2010
Silv
June
2010
Silv
July
2010
Hyslop
March
2010
Hyslop
April
2010
Hyslop
May
2010
Hyslop
June
2010
Hyslop
July
2010
0.999
0.999
0.999
0.999
0.999
0.999
0.999
0.999
0.999
0.999
-0.02
-0.01
-0.02
0.01
-0.01
-0.01
0.01
-0.03
-0.02
-0.03
0.07
0.06
0.08
0.06
0.08
0.06
0.07
0.09
0.09
0.12
0.997
0.995
0.996
0.992
0.997
0.996
0.995
0.997
0.965
0.996
0.13
0.00
-0.36
-0.45
-0.38
-0.08
-0.32
-0.71
-0.99
-1.27
0.16
0.18
0.37
0.996
0.993
0.996
0.11
-0.01
-0.38
0.913
0.63
0.73
0.970
0.65
0.75
0.984
0.84
0.84
0.990
0.70
0.70
0.995
1.08
1.08
0.984
0.32
0.40
0.992
0.50
0.50
0.985
0.79
0.79
0.987
0.65
0.65
0.994
0.67
0.67
0.16
0.20
0.38
V2 and
Actual 0.47
0.992 -0.45
Mean Temp
0.44
0.997a strong
-0.39
0.45
showed
correlation, but no
0.16
0.996 -0.08
0.15
trends from month
to0.994
month.
0.33
-0.30
0.33
0.47
0.71
1.00
1.27
0.997
0.960
0.996
-0.73
-1.01
-1.30
0.73
1.03
1.30
Daily RH Regime
V2 Relative Humidity lags behind Actual across all
seasons and sites.
Relative Humidity
V2 and Std 60,
typically do not
correlate well for
Max RH.
Crop
Grass
HYS
2010
Grass
HYS
2011
Grass
JCTY
2010
Grass
JCTY
2011
Grass
SILV
2010
Grass
SILV
2011
Grass
DTN
2011
Grape
ARSM
2010
Grape
Croft
2010
Grape
THill
2010
Grape
Wren
2010
Daily Max Relative Humidity
Mean
absolute
Mean
differenc
r
difference
e
Daily Min Relative Humidity
r
Mean
differen
ce
Mean
absolute
difference
0.7916
-6.47
6.57
0.8863
-1.97
4.06
0.7512
-6.52
6.62
0.8658
-2.39
4.08
0.5395
-4.49
4.99
0.8475
-2.95
4.85
0.6328
-6.90
6.92
0.9009
-3.59
4.82
0.5823
-3.29
4.56
0.8846
-2.39
5.05
0.5876
-1.67
3.75
0.7435
-5.01
7.70
0.7801
-4.28
4.38
0.8431
0.78
3.58
0.8280
-2.19
4.19
0.8383
4.77
6.96
0.7230
-5.11
5.93
0.9216
1.98
3.99
-0.0469
0.7044
4.95
-4.55
5.07
4.63
0.4922
0.8499
16.00
3.75
24.66
5.82
Typically, V2 and
Std 60 correlate
well for Min RH, but
could be better.
Min Relative Humidity
V2 Min RH usually
closely aligns with Std
60.
At THill, V2 Min RH
varies widely.
Dew Point Temp
Crop
Grape
ARSM
2010
Grape
Croft
2010
Grape
THill
2010
Grape
Wren
2010
Grape
ArSm
2011
Grape
THill
2011
Grape
BPP
2011
Daily Max Dew Point Temp
Mean
Mean
absolute
r
difference difference
Daily Min Dew Point Temp
Mean
Mean
differen
absolute
r
ce
difference
Daily Mean Dew Point Temp
Mean
Mean
differen
absolute
r
ce
difference
0.8497
1.83
2.13
0.7970
3.40
3.45
0.9055
2.19
2.21
0.9176
1.55
1.68
0.7655
4.33
4.34
0.9048
2.62
2.62
0.3907
13.09
13.11
0.7191
3.06
3.46
0.7329
6.00
6.04
0.8413
2.53
2.56
0.8628
4.10
4.10
0.9242
3.05
3.05
0.8545
1.46
1.97
0.7510
4.64
4.76
0.8690
2.58
2.80
0.5312
4.91
4.92
0.7648
3.94
4.12
0.8943
3.69
3.69
0.9454
0.65
1.06
0.8501
4.31
4.32
0.9634
2.18
2.22
Max and Mean Dew Point Temp, correlate well between V2 and Std 60.
The exception is THill. Min Dew Point Temp shows a weaker correlation
between V2 and Std 60.
Daily Leaf Wetness
Hours of Leaf Wetness
Precipitation
V2 overestimates Precipitation
across sites. V2 usually is
accurate on predicting rain
events when they do occur.
HPM Graphs
Downy Mildew Graphs
V2 Downy infection risk values
typically overestimate Actual
infection values.
Grape Powdery Mildew
V2 underestimates GPM risk
values under 100, and like HPM,
occasionally shows early infection
Rust Models
Interface to Disease Maps via MyPest Page - http://uspest.org/risk/models
This project was supported by the Agriculture and Food Research
Initiative Competitive Grants Program No. 2010-85605-2054 from
the National Institute of Food and Agriculture.
Gridded Disease Maps
Gridded data Example 2: PRISM data for Precipitation compared
to Precip/Disease Maps Interface
IPPC Interface & V2 data– Willamette Valley
PRISM Data - 2 Regions
Comparable data for a rainfall event
This project was supported by the Agriculture and Food Research
Initiative Competitive Grants Program No. 2010-85605-2054 from
the National Institute of Food and Agriculture.
Grape bunch rot
disease grid and
overlay on Google
map
Gridded data Example 1: PRISM and IPPC V2 data for
temperature with example statistical comparison
PRISM Data - 4 Regions – ca. 72 hr lag IPPC V2 data – 3 of 4 regions ca. 12 hr la
This project was supported by the
Agriculture and Food Research
Initiative Competitive Grants Program No.
2010-85605-2054 from
the National Institute of Food and
Agriculture.
Rust Models
These are two examples of
how the V2 predicts the
latent period in the Rust
model when there are
differences.
My virtual station
Validated forecasts
Western Weather Work Group Current Work
Supporting
existing and
emerging
systems
through
distributed
resources
Spotted Wing Drosophila
Overwintering Mortality
Area wide IPM coddling moth
Use of Western Weather Workgroup-developed degreeday and phenology models is increasing nationwide
Publications
• Gent, D. H., De Wolf, E. D, and Pethybridge, S. J. 2011. Perceptions of
risk, risk aversion, and barriers to adoption of decision support systems
and IPM: An Introduction. Phytopathology 101:640-643.
• Pfender, W. F., Gent, D. H., Mahaffee, W. F., Coop, L. B., and Fox, A. D.
2011. Decision aids for multiple-decision disease management as
affected by weather input errors. Phytopathology 101:644-653.
• Gent, D. H. , Mahaffee, W. F., McRoberts, N., and Pfender, W. F. 2013. The
use and role of predictive systems in disease management. Annual
Review of Phytopathology. In press.
• Pfender, W. F., Gent, D. H., and Mahaffee, W. F. 2012. Sensitivity of
disease management decision aids to temperature input errors
associated with out-of-canopy and reduced time-resolution
measurements. Plant Disease 96:726-736
• https://www.facebook.com/pages/Northwest-Hop-InformationNetwork/147514331928522
1. Montana State - "SPUD" potato weather network (ingest station data; deliver disease
models) - MSU contact Nina Zidack
2. WSU - deliver 6.5 day Fox Weather LLC/IPPC hourly weather forecasts - WSU contact
Gerrit Hoogenboom
3. WSU - provide 1st incidence of potato late blight Google maps for Columbia Basin WSU contact Dennis A. Johnson
4. UC Davis - ingest station data from multiple weather networks (incl. PESTCAST, ADCON,
and METOS) and link to multiple disease models and provide virtual weather networks
and data for supported wine grape growers - UC Davis contacts Doug Gubler, Brianna
McGuire; UC Cooperative Extension contact Lynn Wunderlich
5. UC Davis - providing virtual weather station networks for grape IPM (same as
preceeding contacts)
6. UC Davis - developed and add new phenology models for western flower thrips and
Asian citrus Psyllid (Asian citrus Psyllid work supported by a SCRI grant). UC Davis contact:
Neil McRoberts
7. APHIS/PPQ/CPHST/Ft. Collins and Aurora, CO (numerous other interested parties) - add
several models to uspest.org over past several years, including: Brown Marmorated Stink
Bug, European grapevine moth, pine shoot beetle, light brown apple moth, Cereal leaf
beetle, Gypsy moth, emerald ash borer. Supplied daily-updated degree-day grids since
2008 (currently used for backup).
8. Wyoming - developed and added a model for Bauer Spring wheat, Contact Wyoming
Extension Service (Sandra Frost)
9. All states - added a new Google maps based interface to run degree-day models, greatly
improving accessibility to our currently supported 80 models, for all US states, especially
those underserved and without statewide weather networks or models.
10. All states - developed a "web services" interface so that any model and weather
station in our system can be specified and run from remote web pages, such as a county
Extension website. A version of this feature is being used by UC Davis for grape IPM.
11. All states - developed "virtual weather data" and implemented to fill-in missing or
flagged-as-suspicious weather data for all stations in our database.
12. All states - developed modified leaf wetness estimations allowing disease risk models
to be run from weather stations that do not have leaf wetness sensors.
Next Steps
• Expand Impact and Adoption
Assessment
• Expand Infrastructure Support
through signature programs
• PRIME
• LAMP
Thank you!
Monthly Min Temp Data
Daily
minimum
temperature
Silv
March
2010
Silv
April
2010
Silv
May
2010
Silv
June
2010
Silv
July
2010
Hyslop
March
2010
Hyslop
April
2010
Hyslop
May
2010
Hyslop
June
2010
Hyslop
July
2010
0.998
0.991
1.000
0.993
0.999
0.997
0.997
0.996
0.996
0.997
0.12
0.22
0.09
0.26
0.23
0.15
0.13
0.13
0.15
0.19
0.12
0.22
0.09
0.26
0.23
0.15
0.13
0.13
0.15
0.19
0.990
0.992
0.993
0.988
0.993
0.992
0.993
0.995
0.941
0.993
0.50
0.54
0.50
0.54
0.986
0.985
0.63
0.75
V2 Monthly Min
0.61
0.61
0.992
0.69
Temp data showed
a0.54
decline0.980in the
0.54
0.80
correlation with
0.91
0.91
0.987
1.14
Std 60 from Spring
to Summer.
It
0.43
0.44
0.990
0.58
appears to occur in
0.44
0.44
0.992
0.57
May.
0.34
0.77
1.14
0.37
0.91
1.14
0.992
0.940
0.990
0.47
0.91
1.33
0.882
1.15
1.25
0.817
0.84
1.19
0.704
1.41
1.43
0.670
0.95
1.11
0.358
2.05
2.15
0.936
0.45
0.74
0.856
0.77
0.90
0.653
1.36
1.47
0.694
0.75
1.06
0.457
1.17
1.91
0.63
0.75
0.69
0.80
1.14
0.59
0.57
0.48
0.99
1.33
Rust Models
V2 is
unpredictable in
its over and
underestimation
of Infection
Values.
Management Recommendations for
Hop Powdery Mildew.
Site
Year
Weather input
b
Mean error in index
Mean absolute error in index
No. of fungicide
a
applications
Application
interval (days)
9
9.7
-4.19
8.17
10
8.7
10
8.7
10
8.7
9
9.7
10
8.7
10
8.7
10
8.7
9
9.7
9
9.7
10
8.7
10
8.7
10
8.7
9
9.7
10
8.7
10
8.7
9.6
9.1
9.8
9.0
11
7.9
11
7.9
12
7.3
12
7.3
11.5
7.6
11.5
7.6
Oregon
Hop
105
2010
In-canopy 15 min
V2
105
2011
In-canopy 15 min
205
2010
In-canopy 15 min
V2
V2
205
2011
2010
2011
2010
2011
-9.19
11.99
0.43
2.26
-2.26
8.49
In-canopy 15 min
V2
Mean
5.22
In-canopy 15 min
V2
405
-1.56
In-canopy 15 min
V2
405
9.95
In-canopy 15 min
V2
305
-2.63
2.47
In-canopy 15 min
V2
305
-1.51
-2.15
2.58
In-canopy 15 min
V2
-2.88
6.39
12.55
17.99
Washington
Hop
H35
2010
In-canopy 15 min
V2
H35
2011
In-canopy 15 min
V2
Mean
-3.33
9.56
In-canopy 15 min
V2
4.61
13.78
V2 typically
called for one
more spray
than the
Canopy 15. The
spray interval
was usually
shorter for the
V2.